Collimation of plasma-produced x-rays by spherical crystals: Ray-tracing simulations and experimental results

REVIEW OF SCIENTIFIC INSTRUMENTS VOLUME 70, NUMBER 3 MARCH 1999 Collimation of plasma-produced x-rays by spherical crystals: Ray-tracing simulations and experimental results M. Sanchez del Rio a) European Synchrotron Radiation Facility, BP 220, F-38043 Grenoble Cedex 9, France M. Fraenkel and A. Zigler Racah Institute of Physics, The Hebrew University of Jerusalem, Jerusalem 91904, Israel A. Ya. Faenov and T. A. Pikuz Multicharged Ions Spectra Data Center of VNIIFTRI, Russian Committee of Standards, Moscow Region, 141570, Russia Received 16 June 1998; accepted for publication 30 November 1998 Ray-tracing simulations, validated by experimental results, demonstrate that high intensity collimated x-ray beams can be produced from an isotropic x-ray source. A spherically bent mica crystal was used to collimate and monochromatize x rays emitted by a femtosecond laser-produced plasma. The result is a short pulse x-ray beam with a high degree of collimation less than 1 mrad divergence, good spectral resolution (10 2 / 10 4 ), and tunability over a wide spectral range. The role of the experimental parameters in the resulting beam divergence is thoroughly analyzed by ray-tracing modeling. These simulations are validated by test experiments. The ray-tracing calculations define a set of boundaries in the experimental parameters, which will guarantee the achievement of collimated beams better than 1 mrad in further experiments. 1999 American Institute of Physics. S0034-6748 99 01303-9 I. INTRODUCTION The delivery of high intensity x rays to the sample is essential in a variety of experimental techniques and applications compositional mapping techniques, x-ray imaging of biological and other structured samples, etc.. During the last years, we have witnessed a revolution in the ability to generate, manipulate, and detect x rays cf. review by Ceglio in Ref. 1. This progress is the result of technological developments that revitalized old ideas and materialized them into practical x-ray devices. Few examples of the recent x-ray technology are high-brightness x-ray sources laser produced plasmas, Z-pinches, plasma focus, x-ray lasers, and synchrotron radiation facilities, new optical components multilayer mirrors and supermirrors, zone plate and phasezone-plate lenses, Bragg Fresnel lenses, capillaries and Kumakhov lenses, curved Bragg and Laue crystals, compound refractive lenses, etc., and advanced detection systems xray sensitive charge coupled device CCD cameras, fluorescence screens, gas-filled detectors, etc.. X-ray lasers 2 offer many advantages for future applications due to their exceptional characteristics: high intensity, high degree of collimation the radiation can be transported long distances without losses, coherence, monochromaticity, and tunability. However, the present x-ray lasers are in a developmental status, require large facilities, have very low efficiencies, and are limited to the soft x-ray spectral range 45 Å. At present, synchrotron radiation facilities are the most reliable source of high brightness x rays. These facilities produce intense x-ray beams with excellent characteristics of spectral range, collimation, polarization, and even coherence, although they require a large and dedicated installation. This article demonstrates the feasibility of a method which produces monochromatic soft x-ray beams with a high degree of beam collimation, and a relative tunability in the wide spectral range of 20 Å. These beams can now be used in relatively small installations using an x-ray source like X-pinch or laser-produced plasma, and optics based on spherical crystals of high quality e.g., quartz or mica 3 7. The influence of various experimental parameters e.g., source size, crystal radius, etc. on the quality of beam collimation is evaluated by ray-tracing modeling. The raytracing method yields a reliable quantitative estimate of spot dimensions, intensity, and spectral resolution of the x-ray beam delivered by an optical system. This method is widely used in the x-ray domain for the design and optimization of synchrotron beamlines 8,9 and astronomical devices. The advantages of this method for modeling x-ray images produced by an x-ray backlightening scheme using spherically bent crystals and plasma x-ray sources have been shown recently in Ref. 10. Ray-tracing calculations are performed in the context of this work to obtain a quantitative description of the effect of several experimental parameters on the resulting x-ray beam divergence. Such a task is extremely complicated if only analytical calculations are applied. The results of the ray-tracing modeling of collimated monochromatic x-ray beam formation by bent crystals are compared with the experimental results. A quasiparallel beam of x rays in the spectral range around 9.5 Å was obtained by using a spherically bent mica crystal that collia Electronic mail: srio@esrf.fr 0034-6748/99/70(3)/1614/7/$15.00 1614 1999 American Institute of Physics

Rev. Sci. Instrum., Vol. 70, No. 3, March 1999 Sanchez del Rio et al. 1615 FIG. 1. Schematic view of the system used for ray-tracing modeling. mated x-ray radiation emitted from a plasma source produced by a table-top femtosecond laser. II. RAY-TRACING SIMULATIONS OF X-RAY BEAM DIVERGENCE We have simulated the performance of a system consisting of an x-ray source and a curved crystal see Fig. 1. In the simulations, the x-ray plasma source was represented by either a point source or a cylindrical source of several diameter and length values with its axis directed along the optical axis y the line between the source center and the crystal pole. We assume that the x-ray plasma source emits isotropically in all directions. In the simulations, for efficiency reasons, the rays are emitted into a reduced solid angle defined by the semi-aperture of the emission cone, usually 0.13 rad and illuminate only a given area of the crystal about 1 1 cm 2, which is placed at 5 cm from the source, and is orientated to the required Bragg angle relative to the incoming beam. The crystal is considered as a mirror-like system, which means that each ray is reflected at the crystal surface following the laws of specular reflection incident ray, surface s normal and reflected ray must lie on the same plane; and reflected angle equals the incident angle. The validity of this approximation is discussed below. The reflectivity coefficients are not taken considered in the process. After the reflection, the beam divergence can be evaluated in two ways. It can be computed by calculating the full-width-at-half-maximum FWHM of the histogram representing the beam divergence distribution as a function of either the x or z direction. Another possibility is to image the x-ray source at two different planes and calculate the divergence from the shift of the rays x,z coordinates from plane to plane. The latter is used in the experimental setup. When a ray arrives at the crystal surface, it is reflected efficiently only if its wavelength fulfills the Bragg condition: 2d sin n, where d is the interplanar spacing of the crystal, is the grazing angle, n is the order of reflection, and is the wavelength. The x-ray plasma source has a wide spectral distribution in the working energy range in our case, around 9.5 Å. The narrow range of wavelengths that will be reflected is determined by the crystal geometry, in other words, by the beam divergence accepted by the crystal. For a point source and a given harmonic, the spectral range of the reflection is 2nd(sin max sin min ), where max and min are the maximal and minimal angles of incidence of all the rays that arrive at the crystal surface. Each point of the crystal surface, for a given incident angle, is exposed to photons with all possible energies. This holds because the x-ray plasma radiation, unlike synchrotron radiation, presents no correlation between emission angle and spectral distribution. Therefore, for a given point in the crystal surface, and for a given incident angle, the only photons that are reflected specularly are those with an energy which verifies the kinematical Bragg law. The direction of the reflected photon is defined by the specular reflection law. Specular reflection is verified in all the rays impinging the crystal surface with angles inside the Darwin width in a specific configuration: symmetrical crystals in the Bragg geometry, which is exactly our case. Other crystal configurations asymmetrical and Laue do present specular reflection, but only for the rays that fulfill the kinematical Bragg law. Other rays impinging on the crystal inside the Darwin width are nonspecularly reflected, with a direction that depends on the incident angle and photon energy dispersive systems. 11 In our case, the exit beam will present a correlation between angle and photon energy, that does not affect the final image unless an additional filtering or dispersive element would be added downstream from the crystal. The final image and beam divergence are independent from the photon energy and energy bandwidth, thus justifying the ray-tracing simulation of the crystal as a simple mirror when the incident beam is white or its spectral bandwidth is large. When illuminating the crystal with a monochromatic beam, or when its spectral bandwidth is of the order of the intrinsic resolving power of the crystal, the mirror-like approximation is no longer valid. In such a case, the intensity decrease due to the diffraction process must be considered. In particular, if the Bragg angle is not close to the normal incidence case, the beam divergence increases for usual optics i.e., mirror-like systems due to astigmatism. However, for crystal optics, the divergence of the diffracted x-ray beam is kept low by the influence of the Darwin curve of the crystal. This case will be addressed in a following article. Ray-tracing calculations were performed using the SHADOW package. 9 Calculations with different system parameters were done to study their respective role in the final beam divergence. The goal was to obtain collimation lower than 1 mrad. We investigated the effects of the surface shape spherical, parabolic, etc., geometrical errors of the crystal figures uniformity of the curvature radius, crystal size or entrance pupil, source depth and size, and grazing Bragg angle. The simulations set boundaries on these parameters, which guarantee the achievement of collimated beam with divergence lower than 1 mrad. A. Effect of the surface shape and geometrical errors The ideal surface shape for collimating a divergence source is a parabola. A point source placed in its focus will be converted to a perfect collimated beam zero divergence after the reflection. Aspherical surfaces are very difficult to manufacture, polish, and align. Therefore, in most cases, spherical surfaces that approximate the ideal aspherical ones are used. In our configuration, the optical system is not cen-

1616 Rev. Sci. Instrum., Vol. 70, No. 3, March 1999 Sanchez del Rio et al. TABLE I. Beam divergence FWHM in mrads produced by different surface shapes spherical, toroidal, and either parabolic Par or ellipsoidal and focal distances of infinity theoretically collimated, 1150 cm theoretically 1 mrad, and 450 cm theoretically 2.4 mrad. The three different focal positions are obtained by changing the crystal radius or related surface parameters, as ellipse semi-axes. For the spherical surface, the considered radius is the tangential one. A point source emitting isotropically into a cone with semi-aperture angle of 0.13 rad is placed at 5 cm from the collimating surface for all cases. Each divergence value has been calculated averaging five runs of 5000 rays. The error limits are estimated from the standard deviation of the results of the five runs. Surface to focus at infinity Surface to focus at 1100 cm Surface to focus at 440 cm B deg Sphere Toroid Par Sphere Toroid Ellipsoid Sphere Toroid Ellipsoid 86x sag 0.14 0.01 0.14 0.02 0 0.15 0.01 0.66 0.07 0.96 0.04 1.05 0.05 2.41 0.10 2.44 0.06 86 z tan 0.35 0.03 0.37 0.01 0 0.13 0.01 0.13 0.01 0.97 0.03 1.26 0.07 1.21 0.06 2.45 0.02 74 x sag 15.9 0.5 0.46 0.04 0 15.0 0.6 0.49 0.05 0.97 0.02 13.8 0.44 0.37 0.11 2.36 0.11 74 z tan 1.78 0.05 1.64 0.06 0 1.67 0.11 1.60 0.05 0.97 0.02 1.48 0.03 1.42 0.06 2.39 0.08 tered. The radius of curvature of the reflecting surface must thus have different values when focusing in the tangential tan plane our diffraction plane or sagittal sag plane. It can be calculated with the following formulas 1/p 1/q 2/ R t sin, 1 1/p 1/q 2 sin /R s, where p is the source-mirror distance around 5 cm in our case, q is the mirror-image distance infinity for collimating devices, R t (R s ) is the tangential sagittal radius, and is the grazing angle the Bragg angle in our case. A spherical surface (R s R t ) will focus in both directions with the same focal distances p,q only if 90 normal incidence. Otherwise, the focal image will lie in different positions for the tangential and sagittal planes astigmatism. In our case, when setting the spherical radius to focus in the tangential diffraction plane, the distortion in the sagittal direction increases when the incident beam separates from normal incidence. A deviation or error in the curvature radius produces a defocus or shift of the image focal position from Eqs. 1. Finding a best focus in a position q is equivalent to consider a beam divergence of A s /q, where A s aperture stop is the acceptance of the crystal due to its finite dimensions. Using A s 1.1 cm corresponding to a source emitting in a cone with semi-aperture angle of 0.13 rad placed at 5 cm from the crystal, and a defocus equivalent of 1 mrad i.e., q 1100 cm, we obtain from Eqs. 1 a deviation in the radius of curvature of approximately 0.04 cm R t 10.02 cm for q ; R t 9.98 cm for q 1100 cm. Ithas been found by ray-tracing that the effect of aberrations may play a positive role: for 86, with a spherical surface, the deviation in curvature radius limiting the divergence to 1 mrad, will be R 0.11 cm R t 9.91 cm for q 440. Results of these calculations are shown in Table I. It is noteworthy that for angles close to normal incidence 86 the spherical crystal approximates very well the ideal surface, and the use of a toroidal crystal would not improve performance. B. Effect of mirror size or entrance pupil A nonideal crystal surface produces negative effects on the reflected beam divergence. They are more important when the mirror accepts a wide solid angle of radiation. The deviation between the spherical surface and the ideal surface is higher when the distance to the mirror pole increases. For the case of 86, the isotropic incident radiation from a point source in a cone with semi-aperture of up to 0.22 rad which in our case covers an area of almost 2 2 cm 2 of the crystal gives collimation values lower than 1 mrad in the reflected beam. In the case of 74, such a collimated beam is only obtained if the accepted angle in the tangential direction is reduced to 0.10 rad the crystal footprint is lower than 1 cm 2. Numerical results from these calculations are given in Table II. C. Effect of source depth and source width The effect of both source depth and source width results in increasing the beam divergence. The formed can be con- TABLE II. Beam divergence FWHM, in mrads produced by a point source emitting isotropically for different values of source solid angle emission the value is the cone semi-aperture in rads. A spherical crystal is used. The values of the footprint on the crystal correspond to a distance source-crystal of 5 cm. The resulting beam divergence depends only on the solid angle accepted. Therefore these results can be applied to any value of p when the crystal radius is changed accordingly from Eqs. 1. Each value has been calculated averaging five runs of 5000 rays. The error limits are estimated from the standard deviation of the results of the five runs. Semi-aperture of emission cone rads 0.012 0.04 0.07 0.10 0.12 0.18 0.25 86 x sag 0.10 0.01 0.31 0.01 0.50 0.02 0.38 0.07 0.14 0.01 0.39 0.06 1.27 0.01 86 z tan 0.03 0.01 0.04 0.01 0.12 0.01 0.22 0.02 0.36 0.03 0.61 0.05 0.97 0.08 Footprint cm 2 0.10 0.1 0.21 0.33 0.59 0.56 0.84 0.80 1.10 1.05 1.53 1.51 2.11 2.12 74 x sag 1.53 0.05 5.06 0.15 8.91 0.28 12.6 0.5 16.0 0.7 21.1 0.9 27.1 1.1 74 z tan.015 001 0.17 0.01 0.51 0.02 1.04 0.01 1.78 0.03 3.40 0.12 6.71 0.21 Footprint cm 2 0.10 0.10 0.35 0.34 0.61 0.58 0.87 0.84 1.12 1.09 1.58 1.51 2.07 2.08

Rev. Sci. Instrum., Vol. 70, No. 3, March 1999 Sanchez del Rio et al. 1617 TABLE III. Beam divergence FWHM in mrads produced by a line source along the y axis. The source is emitting isotropically. Each value has been calculated averaging five runs of 5000 rays. The error limits are estimated from the standard deviation of the results of the five runs. y m 0 25 100 500 1000 2500 5000 86 x sag 0.14 0.01 0.13 0.01 0.16 0.02 0.29 0.04 0.28 0.03 0.58 0.06 1.40 0.26 86 z tan 0.39 0.03 0.36 0.01 0.35 0.01 0.40 0.02 0.49 0.03 0.86 0.09 1.40 0.13 74 x sag 16.0 0.7 15.9 0.5 15.9 0.7 16.0 0.5 15.5 0.4 14.3 0.2 10.6 0.4 74 z tan 1.79 0.03 1.76 0.10 1.75 0.04 1.77 0.05 1.72 0.10 1.68 0.12 2.2 0.1 74 x sag 0.47 0.03 0.45 0.09 0.42 0.11 0.52 0.09 0.50 0.04 0.64 0.06 1.32 0.20 74 z tan 1.73 0.10 1.70 0.04 1.77 0.12 1.67 0.11 1.70 0.12 1.59 0.13 2.14 0.21 sidered as an indetermination in the p value. The latter introduces in the system an additional divergence of source width /p. Using a line source of length y along the y axis with a spherical mirror collecting radiation in a cone with semiaperture of 0.13 rad and Bragg angle of 86, the simulations show that it is possible to increase the source depth up to values of about 3 mm maintaining the resulting beam divergence below 1 mrad. In the case of 74, and when considering only the tangential plane, the diffracted beam which showed a divergence of 1.73 mrad with a point source maintains its divergence almost constant until increasing the source depth to about 3 mm see Table III. For studying the effect of the source width, we simulated a source consisting of a disk with diameter D in the x,z plane and zero depth in z direction. The divergence produced by the spherical crystal with 86 is determined by the aberration effects when the source diameter is lower than 50 m (D/p 10 3 rad). For higher D values, the effect of the source size dominates over the aberrations in defining the resulting beam divergence Table IV. D. Effect of incident angle When the incident beam is close to the normal incidence position, the effect of the aberrations is small and the diffracted beam is well collimated in both the tangential and sagittal plane. This is still the case when 86. When the incident beam on the spherical crystal is not normal as in the 74 case, the beam is less collimated in the tangential plane, and out of collimation in the sagittal plane. Using an isotropic cylindrical source with both diameter and depth values of 25 m to match the experimental conditions, and introducing the mirror curvature radius calculated from Eqs. 1, we obtained collimated beams with divergence of less than 1 mrad in the tangential direction for Bragg angle of up to 80, and in the sagittal direction only up to an angle of 85 Table V. III. TEST EXPERIMENTS WITH FEMTOSECOND TABLE-TOP LASER AND SPHERICALLY BENT CRYSTAL The experimental setup for testing the feasibility to obtain collimated beams is presented in this paragraph. The TABLE IV. Beam divergence FWHM in mrads produced by a circular source in the x,z plane and no depth in y. Source size values D correspond to the diameter value in m. Each value has been calculated averaging five runs of 5000 rays. The error limits are estimated from the standard deviation of the results of the five runs. D m 0 10 25 50 100 86 x sag 0.14 0.01 0.37 0.09 0.63 0.02 0.89 0.04 1.57 0.08 86 z tan 0.39 0.03 0.38 0.03 0.49 003 0.82 0.04 1.64 0.04 74 x sag 16.0 0.7 16.0 0.8 16.4 0.4 16.1 0.6 15.9 0.4 74 z tan 1.79 0.03 1.75 0.05 1.77 0.04 1.85 0.14 2.19 0.10 74 x sag 0.47 0.03 0.67 0.05 0.85 0.04 1.18 0.01 1.78 0.06 74 z tan 1.73 0.10 1.74 0.07 1.71 0.11 1.73 0.09 2.08 0.10

1618 Rev. Sci. Instrum., Vol. 70, No. 3, March 1999 Sanchez del Rio et al. TABLE V. Beam divergence in FWHM in mrads produced by an extended source 25 m diameter and 25 m depth emitting isotropically into a cone of semiaperture angle of 0.13, as a function of the Bragg angle. Each value has been calculated averaging five runs of 5000 rays. The error limits are estimated from the standard deviation of the results of the five runs. The focusing condition may be achieved by either fixing the p distance to 5 cm and changing the crystal curvature to the values in row R or by fixing the curvature radius to the normal incidence radius 10.00 cm and changing the p distance (p (R/2) sin ) as indicated in row p. B deg 90 88 86 84 82 80 74 68 x sag 0.42 0.01 0.40 0.01 0.64 0.05 1.72 0.14 3.54 0.10 5.96 0.13 15.7 0.6 29.8 0.6 z tan 0.42 0.01 0.42 0.01 0.48 0.02 0.67 0.06 0.87 0.11 1.07 0.03 1.78 0.09 2.49 0.21 R cm 10.000 10.006 10.024 10.055 10.098 10.154 10.403 10.785 p cm 5.000 4.997 4.988 4.973 4.9513 4.924 4.806 4.636 x-ray source was a laser produced plasma. The plasma was generated by the interaction of intense femtosecond laser pulses with a solid target. The laser has a pulse width of 120 fs, an energy of about 20 mj per pulse and a repetition rate of 10 Hz. It is based on a Ti:Sapphire oscillator generating 80 fs pulses at 800 nm with a spectral width of about 10 nm. The pulses were amplified by the chirped pulse amplification technique. 12 The pulses were first stretched temporarily to a pulse width of approximately 1 ns and then sent into a regenerative amplifier and a double-pass amplifier. Then, they were recompressed to the pulse width of about 120 fs. The laser was focused on a solid target to a focal spot of approximately 20 m of diameter to produce power density of 5 10 16 W/cm 2. The resulting plasma 13,14 had a density of the order of 10 22 electrons/cm 3, temperature in the range of 100 200 ev and it emitted intense short bursts of x rays. 15 The length of the x-ray pulse in the wavelength range under our consideration 10 Å is of about 1 ps. 16 Other ranges can be obtained by choosing the right target material and laser intensity. The experimental apparatus is shown schematically in Fig. 2. The laser pulse is focused on the samarium target and the x rays are reflected by the spherical crystal. The x-ray plasma source is placed close to the focal point of spherical mirror. The collimated x-ray beam is formed after diffraction FIG. 2. The experimental apparatus. The laser beam is focused on a solid target and the generated x rays are reflected from a flat RbAP spectrometer not shown and a crystal. The collimated beam passed through a metal grid and formed an image of the grid on a film. The films were protected against direct x rays coming from the plasma. by the spherical crystal. As mentioned before, efficient reflections occur for the x-ray spectral range that fulfills the Bragg condition 2d sin n. Because the x-ray source has small dimensions and the radiation arrives at the crystal in a given angular interval, a narrow range of wavelengths, determined by the crystal properties, 5 is reflected. We used a 15 mm z by 50 mm x mica spherically bent crystal mirror the working size was slightly smaller, 12 mm by 46 mm, due to geometrical constraints with interplanar spacing of 2d 19.9 Å and radius of curvature of R c 186 mm focal length F 93 mm. In this configuration, the spectral band / of the reflected beam can go from 1.5 10 4 to 8 10 2, depending on the acceptance size of the crystal. The central wavelength was 9.5 Å and the crystal alignment was done for second order reflection of the mica crystal 0,0,4. A spectrometer using a flat RbAP crystal was set to record the spectral distribution of the plasma radiation in a wide spectral range: 6.5 11.5 Å. The x-ray emission spectrum was measured with the RbAP spectrometer collecting 2400 laser shots Fig. 3 a. The spectrum shows the peak emission around 9.5 10.2 Å, which allowed a simple alignment of the spherical crystals. The emission intensities in Fig. 3 a are calibrated considering the source-film distance, crystal reflection, x-ray window transmission and film response. The total amount of produced x-ray energy in the effective spectral range was estimated as 4 J per pulse. Taking it into account, and including the solid angle occupied by the spherically bent crystal and the crystal peak reflectivity which is of approximately 0.15 for second order of reflection in the relevant angles of incidence range, 17 the total energy reflected from the crystal was 3 nj per pulse. Taking 1 ps for the x-ray pulse duration, leads to a power of 3 kw per shot in the x-ray parallel beam. The divergence of the collimated x-ray beam was measured by using a metal grid placed downstream from the crystal with wires of 170 m thick and a period of 780 m. A photographic film was placed at various distances downstream from the grid. The quality of the generated beam was monitored by measuring the dimensions of the grid image formed by the beam on a photographic film. Fig. 3 b shows the formed image when the grid was placed 10 cm from the film. The quality of the image is conserved at large distances from the grid and even after few thousands of shots. By comparing the images formed at various distances we concluded that the divergence of the beam is around 1 mrad.

Rev. Sci. Instrum., Vol. 70, No. 3, March 1999 Sanchez del Rio et al. 1619 FIG. 3. a Spectrum of x radiation emitted from the samarium plasma around 9.5 Å. 2400 laser shots were collected to form this spectrum. b Image and traces of the grid formed by the collimated x-ray beam. The grid was placed 10 cm from the film. Estimation of the energy per shot made by direct measurement of the films, leads to similar numbers as stated above about 3 nj, 3 kw per pulse. A ray-tracing simulation of the experimental system is shown in Fig. 4. Two x-ray sources of 50 and 200 m of both diameter and depth were used. The Bragg angle was 86 and a bar pattern and an image plate were positioned at the distances shown in Fig. 1. Good agreement with the experimental results was obtained. The method we described for generating intense collimated x-ray beams presents many advantages. Indeed, it can be used for a variety of x-ray experimental techniques. The x-ray source can easily be tuned over a wide spectral range by selecting the laser intensity and target material. The high repetition rate of the table-top lasers allows a continuous experiment, in contrast to the x-ray laser large installations where the repetition rate is much lower 18 few shots per hour FIG. 4. Ray-tracing simulations of patterns produced by an extended source D y 50 m top, D y 200 m bottom, a spherical crystal at 9.3 cm from the source and a grid-pattern as described in Fig. 1. The image is taken 10 cm downstream from the grid-pattern position. The calculated divergences of these patterns were of approximately 0.5 mrad for the system with 50 m source and of 1.7 mrad for the 200 m source. Axes values are in cm. or even per day. The short x-ray pulse duration generated by the ultrashort laser produced plasma thus result in high intensity collimated x-ray beams. For a synchrotron source, the divergence of the x-ray beam is determined by the divergence of the electron beam TABLE VI. Comparison between synchrotron ESRF ID9 beamline 20, x-ray laser based on the Ta XLVI ion 18 and monochromatic collimated x-ray beams from x-pinch plasma, 2 picosecond laser produced plasma, 19 and femtosecond table-top laser produced plasma, described in this article. Type X-pinch X-ray laser Picosecond laser plasma Femtosecond laser plasma Synchrotron X-ray Source Element Cu Ta Ge Sm Wavelength Å 9.9 45 9.22 9.5 1.2 Source size m 50 20 20 100 16 Beam size mm mm 10 45 0.075 0.075 10 30 0.1 12 46 12 1 1 unfocused Monochromaticity / 4 10 3 10 4 3 10 3 1.5 10 4 8 10 2 10 4 Divergence 10 4 rad 5 5 100 200 6 6 8 10 1 x-ray energy J 3.2 10 0.3 0.003 22 Repetition rate pulses per h 1 2 1 2 3 6 36 000 1.3 10 9 Peak power kw 0.3 0.6 50 15 30 3 0.43 Average power W 0.01 0.05 15 30 12 50 30 000 0.002

1620 Rev. Sci. Instrum., Vol. 70, No. 3, March 1999 Sanchez del Rio et al. related to its emittance and the natural divergence of the created photons, which decreases when the energy of both the electron beam and the photon beam increases. In the new third generation synchrotron facilities, the resulting divergence is dominated by the natural divergence. It has a Gaussian shape of mrad 0.3264/E e GeV at the critical photon energy E c kev 0.665E 2 e GeV B T, where E e is the energy of the electrons in the storage ring and B is the magnetic field of the bending or wiggler magnets. For instance, a facility with an electron beam of E e 2 GeV and a magnetic field B 1 T, will produce x-ray beams with a divergence of about 0.4 mrad FWHM at the critical energy of E c 2.7 kev 4.6 Å. A facility like the European Synchrotron Radiation Facility ESRF (E e 6 GeV) will produce beams with approximately 130 rads FWHM divergence at the critical energy of 19 kev. The method presented above can hardly achieve the latter collimation values. However, the source-sample distances can be dramatically reduced with respect to the synchrotron-based system. Table VI summarizes the characteristics of the collimated beams obtained in our experiment, compared to collimated beams obtained by using i x pinch, 3 ii picosecond laser produced plasma, 19 iii the shortest wavelength x-ray laser presently available 18 and iv a x-ray synchrotron source for timeresolved experiments. 20 Comparing x-ray lasers with plasma sources, the x-ray laser presents several advantages like energy per pulse. Conversely, the monochromatic collimated beams produced by the proposed system are better in terms of divergence and wavelength tunability. The femtosecond laser-produced plasma system has the smallest pulse duration, which allows high intensity beams with even low energy per pulse values. The high pulse repetition rate 10 Hz, compared to other plasma sources, enables the collection of a substantial amount of energy in a short time, as indicated in Table VI. Ray-tracing modeling shows see Tables I V that beam divergences of 0.2 0.5 mrad could be obtained experimentally for x-ray source sizes of less than 20 m using bent crystals ideally aligned. The proposed experimental setup can be improved in several ways. The repetition rate can be increased to 100 Hz and even to the KHz region using the new available pump lasers diode-pumped Nd:YAG. Such repetition rates increase the average photon rate by one or two orders of magnitude. Furthermore, new commercially available femtosecond lasers can supply energy of 100 mj per pulse and more, resulting in about 1.2 10 8 x-ray photons/cm 2 per pulse in parallel beams, without change of the geometrical dimension of the laser system. Such intensities can be used for x-ray imaging of chemical and biological samples. The detection can be done by an x-ray CCD camera, using just one x-ray pulse, since the intensity of the parallel beam pulse is higher than the detection edge of such cameras about 5 10 7 photons/cm 2. 21 The monochromaticity of the parallel beam ( / 10 4 10 2 ) is sufficient for such experiments, since the x-ray absorption will vary by less than 0.5% in the spectral range of the parallel beam. 22 The monochromatic x-ray beams obtained with the combination of plasma sources and spherically bent crystals can also be very useful for applications in x-ray backlightening schemes see, e.g., Refs. 23 26. It has been demonstrated, both by experiment and by computer simulations, that it is possible to generate collimated diffracted x-ray beams with divergence of less that 1 mrad using an x-ray plasma source with the sizes of less than 50 m for source-crystal distance of 5 cm. For other source crystal distance values, the source sizes should be scaled appropriately. crystal optics must be set to Bragg angles in the interval 80 90. For obtaining beam divergences inferior to 1 mrad, a smaller source is required. 1 N. M. Ceglio, J. X-Ray Sci. Technol. 1, 7 1981. 2 R. C. Elton, X-Ray Laser Academic, Boston, 1990. 3 A. Ya. Faenov, A. R. Mingaleev, S. A. Pikuz, T. A. Pikuz, V. M. Romanova, I. Yu. Skobelev, and T. A. Shelkovenko, Kvant. Elektron. Moscow 20, 457 1993. 4 T. A. Pikuz et al., X-ray and Extreme Ultraviolet Optics, Proceedings of SPIE, San Diego, 1995 Vol. 2515, p. 468. 5 A. Ya. Faenov et al., Phys. Scr. 50, 333 1994. 6 T. A. Pikuz, A. Ya. Faenov, S. A. Pikuz, V. M. Romanova, and T. A. Shelkovenko, J. X-Ray Sci. Technol. 5, 323 1995. 7 I. Yu. Skobelev, A. Ya. Faenov, B. A. Bryunetkin, V. M. Dyakin, T. A. Pikuz, S. A. Pikuz, V. M. Romanova, and T. A. Shelkovenko, J. Exp. Theor. Phys. 81, 692 1995. 8 M. Sanchez del Rio, Rev. Sci. Instrum. 67, 1996. 9 C. Welnak, G.-J. Chen, and F. Cerrina, Nucl. Instrum. Methods 347, 338 343 1994. 10 M. Sanchez del Rio, A. Ya. Facnov, V. M. Dyakin, T. A. Pikuz, S. A. Pikuz, V. M. Romanova, and T. A. Shelkovenko, Phys. Scr. 55, 735 1997. 11 M. Sanchez del Rio, G. Gruebel, J. Als-Nielsen, and M. Nielsen, Rev. Sci. Instrum. 66, 1934 1995. 12 G. Mourou and D. Umstadter, Phys. Fluids B 4, 2315 1992. 13 F. B. Rosmej et al., J. Quant. Spectrosc. Radiat. Transf. 58, 859 1997. 14 R. Doron, E. Behar, M. Fraenkel, P. Mandelbaum, A. Zigler, J. L. Schwob, A. Ya. Faenov, and T. A. Pikuz, Phys. Rev. A in press. 15 W. H. Goldstein et al., Phys. Rev. E 47, 4349 1993. 16 M. M. Murnane, H. C. Kapteyn, M. D. Rosen, and R. W. Falcone, Science 251, 531 1991. 17 G. Hoelzer et al., Phys. Scr. 57, 301 1998. 18 B. J. MacCowan et al., Phys. Rev. Lett. 65, 420 1990. 19 M. Schnurer, P. V. Nickles, M. P. Kalashnikov, F. Billhardt, A. Ya. Faenov, B. A. Bryunetkin, Proceedings of SPIE 1993, Vol. 2015 p. 261 1993. 20 M. Wulff, F. Schotte, G. Naylor, D. Bourgeois, K. Moffat and G. Mourou, Nucl. Instrum. Methods Phys. Res. A 69 84, 398 1997. 21 Princeton Instruments Catalog on High Performance Digital CCD Cameras September 1995. 22 B. L. Henke, E. M. Gullikson, and J. C. Davis, At. Data Nucl. Data Tables 54, 181 1993. 23 S. A. Pikuz, T. A. Shelkovenko, D. A. Hammer, A. Ya. Faenov, T. A. Pikuz, and V. M. Dyakin, J. Exp. Theor. Phys. 61, 638 1995. 24 S. A. Pikuz, T. A. Shelkovenko, V. M. Romanova, D. A. Hammer, A. Ya. Faenov, T. A. Pikuz, and V. M. Dyakin, Rev. Sci. Instrum. 68, 740 1997. 25 V. M. Dyakin, A. I. Magunov, T. A. Pikuz, I. Yu. Skobelev, and A. Ya. Faenov, Quantum Electron. 26, 75 1997. 26 Y. Aglitskiy et al., Appl. Opt. in press.